WO2011148084A1 - Characterization of the composition of a deposited layer of an alloy of type i-(iii',iii'')-(vi'-vi'') - Google Patents

Abstract

The invention relates to the characterization of the composition of a deposited layer of an alloy of type I-(IH´,III´´)-(Vr,VI") such as for example Cu(In_xGa_{1_x})(S_ySe₁_y)₂. By illuminating the alloy layer with incident radiation of wavelength close to the wavelength corresponding to the forbidden bandwidth of the alloy, a Raman resonance phenomenon is generated. The method then comprises the steps: - measurement of the intensity of a Raman scattering peak characteristic of the alloy of the layer, and determination as a function of said intensity measurement, of a proportion of the III'' species with respect to the III' species, and/or of the VF species with respect to the VI´´ species, in the alloy of the thin layer.

Description

 Characterization of the composition of a deposited layer

 of an alloy of type I- (IirjII ") - (Vr, VI")

The present invention relates to the characterization of a type I - (IH ', IH ") - (Vr, VI") alloy composition, wherein:

 "I" means an atomic species in the first column of the Periodic Table of Elements, such as, for example, copper (or silver or other),

 "ΙΙΓ" means a first atomic species of the third column of the Periodic Table of Elements, such as, for example, indium, gallium or aluminum,

 "III" "means a second atomic species of the third column of the Periodic Table of Elements, such as, for example, indium, gallium or aluminum,

 - "VF" means a first atomic species of the sixth column of the Periodic Table of Elements, such as, for example, sulfur or selenium,

 "VI" means a second atomic species of the sixth column of the Periodic Table of Elements, such as, for example, sulfur or selenium.

This alloy can be obtained typically in the form of a thin layer deposited for example by electrolysis and to which a thermal treatment with addition of VI element (s) was then applied, or else by chemical deposition or by vapor deposition. ("PVD"). It may be, for example, an alloy of the CuIn (S _y Se 1 - _y ) 2 type, or else of the Cu (In _x Ga 1- _x ) (Se) ₂ or Cu (In _x Ga 1- _x ) (S) _{2 type} , or a penternaire alloy of Cu (in _x Gai_ _x) (S _y Sei_ _{y) 2,} wherein x and y are between 0 and 1. the good photovoltaic properties of such materials make their advantageous application in the design of solar panels where the thin layers of such alloys intervene as an active part of the photovoltaic cell. Conventional techniques such as the SIMS technique (for "Secondary Ion Mass Spectroscopy") make it possible to precisely characterize the compositions of such alloys. However, these techniques are destructive. Non-destructive optical characterization techniques, such as, for example, photoluminescence or Raman scattering spectroscopy, are preferred. However, photoluminescence spectroscopy lacks precision. Typically, the defects present in the layer can lead to a half-height enlargement of the photoluminescence peak, making the characterization of the composition imprecise. Moreover, Raman scattering spectroscopy is usually based on the Raman frequency at which a peak of light generated by Raman scattering is detected. However, the fineness of Raman frequency detection of the position of this peak is not sufficient to precisely determine the composition, for this type of alloy in particular. In addition, the Raman frequency position of the Raman scattering peaks can be influenced by other parameters than the composition (stresses, impurities, etc.).

The present invention improves the situation.

It proposes for this purpose to provide an illumination of the alloy layer by an incident radiation wavelength chosen sufficiently close to a wavelength of the bandgap of the alloy to generate a resonance phenomenon Raman (or at least "pre-resonance" Raman).

This results in an overall increase in the intensity of the Raman scattering peaks, and, in particular, a measure of the intensity of at least some of these peaks can finely give the alloy composition, according to an advantage that provides the invention. Indeed, as will be seen in detail below, while the bandgap position of the alloy linearly depends on the alloy composition, the intensity variation of a resonant Raman peak can be expressed in accordance with FIG. exponential dependence depending on the composition. This results in a much finer measurement by resonant Raman scattering than by photoluminescence. Thus, the present invention firstly proposes a method comprising the steps, implemented in the aforementioned resonance (or "pre-resonance") conditions:

 measuring the intensity of a Raman scattering peak characteristic of the alloy of the layer, and

 determine, according to the said intensity measurement, a proportion of the species ΙΙΓ with respect to species III ", and / or of the species VF with respect to species VI", in the alloy of the thin layer.

In one embodiment, it may be chosen to compare this resonant Raman peak intensity measured on the layer with that (s) of one or more standard alloys of well-known compositions, to determine the alloy composition of the layer. .

Thus, in such an embodiment, the method comprises, following the measurement step on the thin layer, the steps:

 comparison with at least one reference measurement of intensity of the same Raman scattering peak on at least one standard alloy of the same type as the alloy of the layer and of known composition, and

 based on this comparison, determine at least one variation (in the sense of a possible deviation), relative to the standard alloy, of a proportion of species ΙΙΓ with respect to species III ", and / or of the species VF with respect to species VI ", in the alloy of the thin layer.

The term "standard alloy of the same type" is used here to mean that the alloying elements of the standard alloy I, ΙΙΓ, III ", VF, VI" are the same as those of the alloy of the layer slim. Of course, the proportion of these elements in the deposited layer may deviate slightly from that desired during the deposition of the layer, and therefore that of a standard alloy that could have this desired composition. For this purpose, a comparison is made with at least this standard alloy. Preferably, the measurement of the intensity of a Raman scattering peak in the layer to be characterized can be compared with several reference measurements obtained. in standard alloys of the same type as the alloy of the layer and of different compositions from one standard alloy to another, by applying the same measurement conditions as on the standard alloys. In practice, the analysis of a Raman measurement on a layer to be characterized will preferably be made by comparison with a calibration curve resulting from the measurement of the intensity of a Raman peak of each standard alloy at a given wavelength. , this calibration curve establishing the dependence of the Raman peak intensity as a function of the alloy composition, as will be seen below with reference to FIG. 6. Thus, preferably, the method comprises a comparison with several measurements of intensity reference of the same Raman scattering peak on respectively several standard alloys of the same type as the alloy of the layer and of known compositions, respectively different. In one possible embodiment, the above-mentioned alloy is a compound of the type CuIn (S _y Sc _y ) 2, with y between 0 and 1. The incident radiation wavelength is then chosen to be adequate to obtain a resonance (or Raman pre-resonance) involving a high intensity of the measured Raman scattering peak. For example, if the alloy composition y is between 0.65 and 1, a wavelength of the incident incident radiation is of the order of 785 nm.

Under these conditions, it was observed that the Raman scattering peak that it was preferable to measure, as its intensity was increased by the resonance phenomenon, was a peak corresponding to an E (L) / B type of vibration mode. ₂ (L) of the alloy, located in this embodiment, at a frequency of the order of 340 - 350 cm ^-1 .

Of course, by increasing the wavelength of the incident radiation, it is possible to change the range of values of the composition y to be determined (for y less than 0.65) and thus cover the entire composition range between y = 0. and y = 1. In a possible variant embodiment, the alloy is a compound of Cu (In _x Ga / _x ) (Se) _{2 type} , with x being between 0 and 1. Here again, the incident radiation wavelength is chosen to obtain Raman resonance (or pre-resonance) resulting in a high intensity of the measured Raman scattering peak.

For a composition x less than 0.6, a radiation wavelength of the order of 785 nm may be suitable, in one exemplary embodiment. For a composition x greater than 0.6, a radiation wavelength rather of the order of 930 nm is preferable.

Of course, the present invention applies equally well to the compound of the type Cu (In _x Ga 1 _x ) (S) ₂ , with x being between 0 and 1, the incident radiation wavelength being chosen (as it is will be discussed later with reference to FIG. 2) to obtain Raman resonance or pre-resonance resulting in a high intensity of said measured Raman scattering peak.

In all cases of alloy compositions, it may be advantageous to further measure the wavelength position of a maximum photoluminescence intensity to determine the band gap width of the alloy and hence complete the characterization of the alloy composition. For this purpose, it is preferable that the wavelength of the incident radiation is less than a wavelength corresponding to the bandgap width of the alloy to simultaneously measure the maximum intensity of photoluminescence.

Thus, the wavelength of the incident radiation is preferably:

less than the wavelength corresponding to the bandgap width of the alloy (in order to simultaneously measure the peak of photo-luminescence of the exciton), but nevertheless chosen sufficiently close to the forbidden bandwidth to generate a Raman resonance or pre-resonance.

Of course, it is possible to cross-check or complete the characterization of the alloy composition by additional measurements of the Raman frequencies and / or the bandgap of the alloy. Thus, the method according to the invention may further comprise a measurement of a wavelength position of a maximum of photoluminescence intensity to determine the energy of the exciton which is close to the width of the forbidden band of the alloy and hence eventually complete the characterization of the alloy composition. It is then possible to determine by these various means all the parameters x, y of the alloy composition, even pen- ternary, of the type Cu (In _x Ga 1- _x ) (S _y Se 1- _y ) 2 (with x and y between 0 and 1).

Thus, the present invention also relates to the characterization compositions x and y of a penternaire alloy of Cu (In _x Gai_ _x) (S _y Sei_ _y) 2, with x and y between 0 and 1.

The present invention also aims at a device for implementing the above method, such a device comprising:

 a laser source,

 a Raman probe (including, for example, a spectrophotometer as will be seen hereinafter with reference to FIG. 1), and

 calculating means (comprising for example a computer connected to the spectrophotometer) for

 measure the intensity of a Raman scattering peak characteristic of the alloy of the layer, and

determine, according to the said intensity measurement, a proportion of the species ΙΙΓ with respect to species III ", and / or of the species VF with respect to species VI", in the alloy of the thin layer. Here, the term "spectrophotometer" any device capable of measuring a quantity of photons received at a given wavelength (such as for example a spectrometer coupled to a photomultiplier). The laser source is preferably a tunable laser of the Titanium-Sapphire type between 700 and 1030 nm, to cover the incident radiation wavelengths stated above.

The present invention also relates to a computer program comprising instructions for implementing, when this program is executed by a processor, the steps of the above method as follows:

 measure the intensity of a Raman scattering peak characteristic of the alloy of the layer, and

 determine, according to said intensity measurement, a proportion of the species ΙΙΓ with respect to species III ", and / or of the species VF with respect to the species

VI ", in the alloy of the thin layer.

Thus, the aforementioned calculation means may be in the form of a computer executing the computer program within the meaning of the invention.

For example, a memory storing in database form the reference values of the resonant Raman peaks for different alloy compositions and different values of incident radiation wavelength (measured initially on standard alloys whose composition has been determined by SIMS for example).

Other features and advantages of the invention will appear on examining the detailed description below, and the attached drawings in which:

schematically illustrates a device for implementing the method within the meaning of the invention; FIG. 2 illustrates the different bandgap widths (or "gap", also referred to hereinafter by the conventional abbreviation "Eg") as a function of the alloy compositions for different alloys of the Cu (In _x Gai_ _x ) system (S _y Sei_ _y ) ₂ , in agreement with the literature, and possible values of incident radiation wavelength (corresponding to commercially available lasers such as a titanium-sapphire laser) for the characterization of alloys in composition;

FIG. 3 illustrates different photoluminescence peaks with superstructures corresponding to resonant Raman scattering peaks (in particular for y = 1) in the compound CuIn (S _y Sc _y ) 2, for different composition values y (between 0 , 7 and 1), the wavelength of the incident radiation being 785 nm here;

FIG. 4 illustrates the variation of the exciton energy PLmax (close to the gap Eg and given here by the photoluminescence maximum measured on spectra such as those represented in FIG. 3) as a function of the composition y (given as a percentage here, between 65% and 100%) in the compound CuIn (S _y Sei_ 5 shows a net increase in Raman scattering peaks (de- convolved by the respective peaks photoluminescence) in the compound CuIn (S _y Sei_ _y) 2, for different composition values y (between 0.7 and 1), at the incident wavelength of 785 nm;

FIG. 6 illustrates the intensity variation of the E (L) / B ₂ (L) mode in the Cu-In-S-Se alloy, as a function of the sulfur content [S] at the same wavelength. 785 nm; and

FIGS. 7 and 8 illustrate the variation of intensity predicted for a pre-resonance mode in the Cu-In _x -Ga ₍ i- _x) -Se ₂ alloy as a function of the indium x content, respectively for a wavelength. incidental 785 nm for Figure 7 and 930 nm for Figure 8, these figures thus illustrating the possibility of extending the range of determinable compositions as a function of the wavelength of the incident radiation. Reference is first made to FIG. 1, on which is shown by way of example a possible device for carrying out the method within the meaning of the invention. In the embodiment shown in FIG. 1, the device comprises:

 a laser source 10, for example a Titanium-Sapphire laser (bearing the reference LAS in FIG. 1), tunable to emit incident radiation whose wavelength is suitable for the range of alloy compositions to be measured, this length of the wave being for example between 700 nm and 1030 nm; a Raman probe 11, described in detail below (comprising for example a spectrophotometer for capturing Raman spectra), and

 computing means 12, for example for:

o identify the maximum intensity of a Raman peak in a predetermined Raman frequency range (in cm ^-1 ), for example around 340-350 cm ^-1 in an embodiment described below, this correspondingly corresponding peak general to a characteristic vibrational mode of the analyzed alloy, and

 o compare this intensity with one or more reference measurements made on the same Raman peak of one or more standard alloys of the same type as the alloy of the analyzed layer but of well-known respective compositions, to determine the variations in proportion of species ΙΙΓ (versus III ") or species VF (versus VI") in the alloy of the thin layer, as will be seen in many embodiments below.

The term "standard alloy of the same type as the alloy of the analyzed layer" is understood here to mean the fact that the layer has been deposited with an expected composition and that the standard alloy with which the Raman intensity is compared is composition close to that expected. Typically, if the measured Raman intensity is very far from that of the standard, while a photoluminescence analysis of the layer shows that the composition should be close to the standard, then this observation can be used to deduce the presence defects in the layer (inhomogeneities, impurities, dislocations, constraints, or other). Preferably, several standard alloys for establishing a calibration curve typically giving the amplitude of a Raman peak as a function of the composition of the standard alloys, for example for a given incident wavelength, as will be seen hereinafter with reference to FIGS. , 7 and 8.

The Raman probe comprises, at the output of the laser LAS, a first optical fiber FOI for coupling the output of the laser LAS to a lens L focusing the beam from the laser on the CM layer whose composition is to be characterized. In the example shown, a bandpass filter F1 is advantageously provided to eliminate any Raman scattering components induced by the FOI fiber.

Objective L also collects the scattered light (Raman scattering, Rayleigh scattering, etc.), resulting from photoluminescence, and possibly reflected, and an SR beam splitter (for example a semi-reflective plate) directs the beam from the beam. diffusion, photoluminescence and / or reflection to a spectrophotometer SPEC. It is also advantageously provided (for example at the output of the beam splitter SR) an additional filter F2 attenuating the beam from the diffusion and / or reflection at the wavelength of the incident laser beam, to limit the intensity of the beam. light sent to the SPEC spectrophotometer at this wavelength. This light corresponds essentially to reflected light or from Rayleigh scattering at the same wavelength as the incident light. This precaution makes it possible to clearly distinguish the Raman scattering peaks from a background noise, as will be seen below with reference to FIGS. 3 and 5. In the example represented in FIG. 1, a second optical fiber is furthermore provided. F02 for routing the reflected and / or scattered light to the spectrophotometer SPEC (with, of course, a filter of any Raman scattering components specific to this fiber F02).

The spectrophotometer SPEC is then connected to a computer 12 (for example a computer equipped with a processor and a working memory) which, for example, can advantageously automatically locate the relevant Raman scattering peaks for the characterization of the spectrophotometer. composition of the alloy in the thin layer CM, determine the intensity of these peaks, to compare with one or more references, and deduce the aforementioned composition. For this purpose, in the example shown, these calculation means 12 comprise a DETEC peak detection module, a COMP module leading to at least one comparison to one or more selected references, and a module (referenced x, y) determining the x, y composition of the alloy according to this comparison.

FIG. 2 shows the different gap values Eg (in eV on the left and in wavelengths, on the right) as a function of the x, y compositions of the penenary Cu (In _x Gai_ _x ) (S _y Sei_y) ₂ (the composition x being expressed by the abscissa of this diagram, while the composition is expressed in a variation along the ordinates). It will then be understood that a Titanium-Sapphire LAS (Ti-Sa) laser, tunable between 700 nm and 1030 nm (hatched region), is a good candidate for the investigation of the x, y composition of the alloy based on a Raman scattering measurement under resonance conditions of the photon-phonon interaction, as explained below with reference to FIG.

There is shown in Figure 3, by way of example, different Raman spectra and photoluminescence (reference PL) for various compositions of the Cu-In-alloy (S _y Sei_ _{y) 2} (with a sulfur content [S] ranging from 70 to 100%, thus for between 0.7 and 1), obtained with a wavelength of incident radiation of 785 nm. Each spectrum shows in particular:

a Rayleigh scattering peak R at the same wavelength as the incident radiation (the Raman frequency shift being then 0 cm ^-1 ),

 a larger peak of photoluminescence PL, whose position of the maximum also makes it possible to characterize the composition y, as will be seen below with reference to FIG. 4,

first-order Raman scattering peaks corresponding to a mode of vibration characteristic of the alloy (for example the E (L) / B ₂ (L) mode for the peak P whose amplitude is remarkable), o and second order Raman scattering peaks 2ND (incident photon interactions with two phonons).

It appears in particular that the Raman signal, for example of the peak P, is all the more intense as the composition is high (more sulfur S than selenium Se). This observation is explained by the fact that the wavelength of the gap Eg is close to the wavelength of the incident radiation (785 nm here), to the Raman frequencies close to 0 cm ^-1 in FIG. therefore when the gap Eg increases in eV in accordance with the left-hand part of Figure 2 towards CuInS ₂ (since CuInSe ₂ ) - and the photon-phonon interaction is more resonant Thus, the peak P of the spectrum obtained for the ternary CuInS ₂ is more intense than that of the spectrum obtained for the compound Cu-In- (S _y Sc _y ) ₂ with y = 0.7, for example Figure 4 confirms the dependence of the position of the maximum in photoluminescence PLmax as a function of the composition y in the alloy Cu-In- (S _y Sei_ _y ) ₂ , but it will be understood that the intensity of the Raman peaks is then also an alternative parameter for the evaluation of the composition y.

FIG. 5 effectively shows different Raman spectra for different sulfur compositions [S]. A marked variation in the amplitude of the Raman peaks and in particular of the peak P related to the E (L) / B ₂ (L) mode is observed at about 340-350 cm ^-1 , depending on the composition [S] (of 68.5% to 100%).

FIG. 6 shows in particular the variation in intensity of this peak P as a function of the sulfur composition y. The statistical correlation R2 based on the study of the intensity of the peak P according to FIG. 6 seems much better than that on the photoluminescence shown in FIG. 4, thus making this Raman intensity a parameter of choice for determining the composition.

However, as noted earlier, other factors may influence Raman peak amplitude. These are defects such as stresses in the material, inhomogeneities, impurities, etc. These defects may help to modify the amplitude of the resonant Raman peaks. Measuring the height of at least one Raman peak (such as the peak P), under resonance conditions, although presenting then an attractive alternative to the characterization of the composition by study of the photo luminescence, can nevertheless rely usefully on a complementary analysis of the photoluminescence for characterization of the composition and / or defects. Moreover, the measurement of the Raman frequency of the peaks (in cm ^-1 ) may be a further useful analysis to characterize the alloy of the layer (and in particular its composition).

In general, the intensity of the Raman peaks may also be dependent on the crystallographic orientation of the CM thin layer if it is completely crystalline. However, this observation (widespread in layer growth by epitaxy) is less sensitive:

 on the one hand, in the case of a polycrystalline layer with variable crystallographic orientations, as in layers typically intended for photovoltaic devices such as alloy layers I-III-VI, and on the other hand, under Raman resonance conditions (in which the intensity of the resonant Raman peak is much more dependent on the excitation energy than on the crystallographic orientation).

Of course, a similar approach as that described above with reference to FIG. 6 can be carried out to determine the composition x in the alloy Cu (In _x Ga 1 _x ) Se ₂ . FIG. 7 shows the dependence of the intensity of a Raman peak under pre-resonance conditions on the composition x for an incident radiation wavelength of 785 nm (favorable to resonance for a range of compositions x ranging from 0 to 0.6). However, for larger values of the composition x (thus for more indium than gallium, going towards the smaller gaps Eg and approaching the alloy CuInSe ₂ starting from CuGaSe ₂ - from the right to the left of FIG. 3), the resonance of the photon-phonon interaction is gradually lost and it is advisable to choose a wavelength of the higher incident radiation (thus a smaller excitation energy in eV to approach the values of the CuInSe ₂ alloy gap, for example 930 nm. The characterization of the composition x, for x between 0.6 and 1, is given at this excitation wavelength of 930 nm in FIG. 8. Again, it will be understood that the intensity of the peak related to the resonant Raman mode makes it possible to characterize the composition x. Similarly, similar results (at appropriate incident radiation wavelengths) can be obtained on the compound Cu (In _x Ga 1 _x ) S 2.

More generally, it has been described in detail above the characterization of the quaternary compound composition from the study of the intensity of Raman scattering peaks only. Of course, the characterization of the composition of penenary compounds (or possibly other more complex alloys in compositions x, y, z, etc.) can also be carried out by complementary determination of the position of the gap and / or the Raman frequencies of the Raman scattering peaks. For example, the invention can be applied also for Cu-In-Al-Ga-S-Se alloys, where Al is the aluminum element.

It will generally be noted that for the characterization of the composition of the alloys I-III-VI in particular (quaternary or penenary), the analysis of the amplitudes of the Raman scattering peaks, under resonance conditions of the photon-phonon interaction , has proved particularly advantageous.

The present invention is not limited to the embodiments described above by way of example; it extends to other variants.

For example, the optical and analysis arrangement of Raman and photoluminescence spectra shown in Fig. 1 are susceptible to variations (e.g. by removing fibers and filters, or providing for Brewster incidence mounting to minimize parasitic reflection of Raman signals, or others). Furthermore, at least one reference measurement on a standard alloy has been described above to determine the composition, by comparison between the measurement on the layer and the reference measurement. Alternatives are nevertheless possible: for example, a comparative measurement of the same peak at a different wavelength (not causing a resonance effect) or at another peak of the same spectrum, not (or less) concerned by the resonance effect.

Claims

1. A method for characterizing the composition of a deposited layer of a type I - (IH ', IH ") - (Vr, VI") alloy, in which:  - "I" means an atomic species in the first column of the Periodic Table of the Elements,  "ΙΙΓ" means a first atomic species in the third column of the Periodic Table of the Elements,  "III" means a second atomic species in the third column of the Periodic Table of the Elements,  "VF" means a first atomic species in the sixth column of the Periodic Table of Elements,  "VI" "denotes a second atomic species of the sixth column of the periodic table of elements, characterized in that the process comprises, during the illumination of the alloy layer by an incident radiation of wavelength chosen close to a wavelength of the bandgap of the alloy to generate a Raman resonance phenomenon, the steps:  measuring the intensity of a Raman scattering peak characteristic of the alloy of the layer, and  determination according to said intensity measurement, a proportion of the species ΙΙΓ with respect to the species III ", and / or the species VF with respect to the species VI", in the alloy of the thin layer. 2. Method according to claim 1, characterized in that it comprises, following the measuring step, the steps: comparison with at least one reference measurement of intensity of the same Raman scattering peak on at least one standard alloy of the same type as the alloy of the layer and of known composition, and depending on the comparison, determine at least one variation, relative to the standard alloy, of a proportion of the species ΙΙΓ with respect to the species III ", and / or the species VF with respect to the species VI "in the alloy of the thin layer. 3. Method according to claim 2, characterized in that it comprises a comparison with several intensity reference measurements of the same Raman scattering peak on respectively several standard alloys of the same type as the alloy of the layer and known compositions. respectively different. 4. Method according to one of the preceding claims, in which the alloy is a compound of CuIn type (S _y Sei_ _y ) 2, with y between 0 and 1, characterized in that said incident radiation wavelength is suitable for obtaining a Raman resonance or pre-resonance involving a high intensity of said measured Raman scattering peak. 5. The method of claim 4, wherein the alloy composition y is between 0.65 and 1, characterized in that the wavelength of the incident radiation is of the order of 785 nm. 6. Method according to one of claims 4 and 5, characterized in that the Raman scattering peak corresponds to a type of vibration mode E (L) / B ₂ (L) of the alloy. 7. Method according to claim 6, characterized in that the Raman scattering peak is located at a frequency of the order of 340 - 350 cm ^-1 . 8. Method according to one of claims 1 to 3, wherein the alloy is a compound of Cu (In _x Ga / _x ) (Se) ₂ , with x between 0 and 1, characterized in that said length of incident radiation wave is adequate to obtain a Raman resonance or pre-resonance involving a high intensity of said measured Raman scattering peak. 9. Method according to claim 8, characterized in that the radiation wavelength is of the order of:  785 nm for a composition x less than 0.6, and  of the order of 930 nm for a composition x greater than 0.6. 10. Method according to one of claims 1 to 3, wherein the alloy is a compound of Cu (In _x Ga / _x ) (S) ₂ , with x between 0 and 1, characterized in that said length of incident radiation wave is adequate to obtain a Raman resonance or pre-resonance involving a high intensity of said measured Raman scattering peak. 11. Method according to one of the preceding claims, characterized in that it further comprises a measurement of a wavelength position of a maximum photoluminescence intensity to determine the bandgap width of the alloy and hence complete the characterization of the alloy composition. 12. The method of claim 11, characterized in that the wavelength of the incident radiation is less than a wavelength corresponding to the bandgap width of the alloy to simultaneously measure the maximum intensity of photoluminescence. 13. Method according to one of the preceding claims, characterized in that the alloy is penenary type Cu (In _x Gai_ _x ) (S _y Sei_ _y ) ₂ , with x and y between 0 and 1. 14. Device for implementing the method according to one of the preceding claims, characterized in that it comprises:  a laser source (10),  a Raman probe (11), and calculation means (12) for o measure the intensity of a Raman scattering peak characteristic of the alloy of the layer, and  o determine, according to the said intensity measurement, a proportion of species ΙΙΓ with respect to species III ", and / or species VF with respect to species VI", in the alloy of the thin layer. 15. Device according to claim 13, characterized in that the laser source is titanium-Sapphire type tunable between 700 and 1030 nm. Computer program comprising instructions for implementing, when this program is executed by a processor, the steps of the method according to one of claims 1 to 12:  measure the intensity of a Raman scattering peak characteristic of the alloy of the layer, and  determine, according to the said intensity measurement, a proportion of the species ΙΙΓ with respect to species III ", and / or of the species VF with respect to species VI", in the alloy of the thin layer.